Prognostics and Health Management using Artificial Intelligence

Abstract

Purpose: This study aims to examine artificial intelligence (AI) algorithms used in prognostics and health management (PHM), present a C-MAPPS data case study from the PHM competition, and identify the further research needed in AI-based PHM.BRMethod: AI algorithms that are widely used in PHM are machine learning (ML) methodologies such as ANN, SVM, DT, and kNN. In this study, we briefly introduced these methods and applied them to the C-MAPPS data presented in the PHM competition.BRResults: An organized series of procedures is needed to utilize AI in PHM. This study presented the proper procedure for using AI in PHM, explaining the necessity of training it with data using ML methodologies and verifying it with verification data.BRConclusion: The recent development of ML has been spreading PHM across industries, coupled with condition-based maintenance, one of the reliability strategies. In this study, we applied ML methodologies, including kNN, SVM, random forest, and LSTM, to the C-MAPPS data and discovered that kNN showed slightly better results than the other models in the data presented.